368 
T. H. GOODSPEED AND R. E. CLAUSEN 
same plants (or their asexual offspring) during two successive seasons 
and similar results were obtained on each occasion. Although these 
studies were primarily concerned with correlation phenomena, the 
data show the same sort of differences which we have pointed out for 
different characters in Nicotiana. For flowers measured early in the 
season as reported by from these data by Weldon (1901) the mean 
number of stamens was 26.73 zb .15 as contrasted with a mean number 
late in the season of 17.86 db .12, a difference of 8.87 =1= .19, which is 
certainly a significant difference. Likewise an early mean number of 
pistils of 1745 d= .16 decreases to 12.15 =t -i^, a difference in this 
case of 5.30 d= .20. If we, likewise, calculate the coefficient of varia- 
bility for these series we find that it increases in the case of number 
of stamens from 14.01 d= .41 early in the season to 18.47 d= .46 late in 
the season, an increase of 4.46 zt .62 per cent. Similarly for number 
of pistils the increase is from 22.32 db .65 per cent to 27.88 =b .69 
per cent, an increase of 5.56 d= .95. Miss Alice Lee (1902) in connec- 
tion with a study of data of several series of plants of Ficaria verna 
Huds. states that ''these changes while demonstrating the fact that 
the four series are not random samples of the same population made a 
the same time, are not by any means greater than the same plant in 
the same locality at different periods of its season or the same plant 
in different districts at the same period has been known to give. They 
are well within the limits of local environmental or seasonal changes." 
Although this is not quite the same character of change as the one with 
which we are dealing, it nevertheless seems desirable to point out 
the fact that wide variations in floral organs are known which can 
have no important genetic significance. 
Our results then go to show that under the stress of external and 
more or less artificial conditions attending development, flower size 
in Nicotiana is subject to distinct and significant variations. Our 
principal interest in this question of the stability of the character 
complex, corolla size, has been centered on the effects produced by the 
internal, inherent, and normal ontogenetic changes that make up 
practically all of what we see in the development of a plant. 
A varying period of vegetative development, more or less constant 
for a given species, precedes the appearance of the floral organs. One 
seems justified in assuming, at least, that the products of metabolic 
activity are devoted during the earliest period of development to the 
establishment of an adequate system of food supply at the expense of 
